# Research Papers — 2025-12 · bundle page 44 of 175 > Full-text markdown of papers #2151–2200 of 8703 published in 2025-12, newest first. AI-readable mirrors of the full 8703-paper 2025-12 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2025-12/page/44 [← Previous bundle](https://aipod.dev/papers/archive/2025-12/page/43.md) [Next bundle →](https://aipod.dev/papers/archive/2025-12/page/45.md) ### A Composable Channel-Adaptive Architecture for Seizure Classification Authors: Francesco Carzaniga, Michael Hersche, Kaspar Schindler, Abbas Rahimi Published: 2025-12-22 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.19123v1 Objective: We develop a channel-adaptive (CA) architecture that seamlessly processes multi-variate time-series with an arbitrary number of channels, and in particular intracranial electroencephalography (iEEG) recordings. Methods: Our CA architecture first processes the iEEG signal using state-of-the-art models applied to each single channel independently. The resulting features are then fused using a vector-symbolic algorithm which reconstructs the spatial relationship using a trainable scalar per channel. Finally, the fused features are accumulated in a long-term memory of up to 2 minutes to perform the classification. Each CA-model can then be pre-trained on a large corpus of iEEG recordings from multiple heterogeneous subjects. The pre-trained model is personalized to each subject via a quick fine-tuning routine, which uses equal or lower amounts of data compared to existing state-of-the-art models, but requiring only 1/5 of the time. Results: We evaluate our CA-models on a seizure detection task both on a short-term (~20 hours) and a long-term (~2500 hours) dataset. In particular, our CA-EEGWaveNet is trained on a single seizure of the tested subject, while the baseline EEGWaveNet is trained on all but one. Even in this challenging scenario, our CA-EEGWaveNet surpasses the baseline in median F1-score (0.78 vs 0.76). Similarly, CA-EEGNet based on EEGNet, also surpasses its baseline in median F1-score (0.79 vs 0.74). Conclusion and significance: Our CA-model addresses two issues: first, it is channel-adaptive and can therefore be trained across heterogeneous subjects without loss of performance; second, it increases the effective temporal context size to a clinically-relevant length. Therefore, our model is a drop-in replacement for existing models, bringing better characteristics and performance across the board. --- ### BanglaForge: LLM Collaboration with Self-Refinement for Bangla Code Generation Authors: Mahir Labib Dihan, Sadif Ahmed, Md Nafiu Rahman Published: 2025-12-22 Categories: cs.SE, cs.CL arXiv: https://arxiv.org/abs/2512.19122v1 Bangla is a low-resource language for code generation, lacking large-scale annotated datasets and tools to transform natural language specifications into executable programs. This makes Bangla-to-code generation a challenging task requiring innovative solutions. To address this, we introduce BanglaForge, a novel framework for generating code from Bangla function descriptions. BanglaForge leverages a retrieval-augmented dual-model collaboration paradigm with self-refinement, combining in-context learning, llm-based translation, systematic prompt engineering, and iterative self-refinement based on execution feedback, where a coder generates initial solutions and a reviewer enhances them for robustness. On the BLP-2025 Bangla Code Generation benchmark, BanglaForge achieves a competitive Pass@1 accuracy of 84.00%, demonstrating the effectiveness of retrieval, model collaboration, and self-refinement for low-resource Bangla code generation. --- ### Stop saying LLM: Large Discourse Models (LDM) and Artificial Discursive Agent (ADA)? Authors: Amar Lakel Published: 2025-12-22 Categories: cs.CL arXiv: https://arxiv.org/abs/2512.19117v2 This paper proposes an epistemological shift in the analysis of large generative models, replacing the category ''Large Language Models'' (LLM) with that of ''Large Discourse Models'' (LDM), and then with that of Artificial Discursive Agent (ADA). The theoretical framework is based on an ontological triad distinguishing three regulatory instances: the apprehension of the phenomenal regularities of the referential world, the structuring of embodied cognition, and the structural-linguistic sedimentation of the utterance within a socio-historical context. LDMs, operating on the product of these three instances (the document), model the discursive projection of a portion of human experience reified by the learning corpus. The proposed program aims to replace the ''fascination/fear'' dichotomy with public trials and procedures that make the place, uses, and limits of artificial discursive agents in contemporary social space decipherable, situating this approach within a perspective of governance and co-regulation involving the State, industry, civil society, and academia. --- ### Generative Giants, Retrieval Weaklings: Why do Multimodal Large Language Models Fail at Multimodal Retrieval? Authors: Hengyi Feng, Zeang Sheng, Meiyi Qiang, Yang Li, Wentao Zhang Published: 2025-12-22 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.19115v2 Despite the remarkable success of multimodal large language models (MLLMs) in generative tasks, we observe that they exhibit a counterintuitive deficiency in the zero-shot multimodal retrieval task. In this work, we investigate the underlying mechanisms that hinder MLLMs from being effective retrievers. With the help of sparse autoencoders (SAEs), we decompose MLLM output representations into interpretable semantic concepts to probe their intrinsic behavior. Our analysis reveals that the representation space of MLLMs is overwhelmingly dominated by textual semantics; and the visual semantics essential for multimodal retrieval only constitute a small portion. We find that this imbalance is compounded by the heavy focus of MLLMs on bridging image-text modalities, which facilitates generation but homogenizes embeddings and finally diminishes the discriminative power required for multimodal retrieval. We further discover that the specific feature components that contribute most to the similarity computations of MLLMs are actually distractors that greatly reduce retrieval performance. Building on these insights, we propose ReAlign, a test-time adaptation approach that applies a whitening transformation to adjust the geometry of MLLM representation spaces. Empirical results show that this simple intervention consistently improves zero-shot multimodal retrieval performance across diverse MLLMs without fine-tuning efforts. The code is available at https://github.com/Heinz217/mllm-retrieval-analysis. --- ### HyperLoad: A Cross-Modality Enhanced Large Language Model-Based Framework for Green Data Center Cooling Load Prediction Authors: Haoyu Jiang, Boan Qu, Junjie Zhu, Fanjie Zeng, Xiaojie Lin, Wei Zhong Published: 2025-12-22 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2512.19114v1 The rapid growth of artificial intelligence is exponentially escalating computational demand, inflating data center energy use and carbon emissions, and spurring rapid deployment of green data centers to relieve resource and environmental stress. Achieving sub-minute orchestration of renewables, storage, and loads, while minimizing PUE and lifecycle carbon intensity, hinges on accurate load forecasting. However, existing methods struggle to address small-sample scenarios caused by cold start, load distortion, multi-source data fragmentation, and distribution shifts in green data centers. We introduce HyperLoad, a cross-modality framework that exploits pre-trained large language models (LLMs) to overcome data scarcity. In the Cross-Modality Knowledge Alignment phase, textual priors and time-series data are mapped to a common latent space, maximizing the utility of prior knowledge. In the Multi-Scale Feature Modeling phase, domain-aligned priors are injected through adaptive prefix-tuning, enabling rapid scenario adaptation, while an Enhanced Global Interaction Attention mechanism captures cross-device temporal dependencies. The public DCData dataset is released for benchmarking. Under both data sufficient and data scarce settings, HyperLoad consistently surpasses state-of-the-art (SOTA) baselines, demonstrating its practicality for sustainable green data center management. --- ### Trifocal Tensor and Relative Pose Estimation with Known Vertical Direction Authors: Tao Li, Zhenbao Yu, Banglei Guan, Jianli Han, Weimin Lv, Friedrich Fraundorfer Published: 2025-12-22 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.19110v1 This work presents two novel solvers for estimating the relative poses among views with known vertical directions. The vertical directions of camera views can be easily obtained using inertial measurement units (IMUs) which have been widely used in autonomous vehicles, mobile phones, and unmanned aerial vehicles (UAVs). Given the known vertical directions, our lgorithms only need to solve for two rotation angles and two translation vectors. In this paper, a linear closed-form solution has been described, requiring only four point correspondences in three views. We also propose a minimal solution with three point correspondences using the latest Gröbner basis solver. Since the proposed methods require fewer point correspondences, they can be efficiently applied within the RANSAC framework for outliers removal and pose estimation in visual odometry. The proposed method has been tested on both synthetic data and real-world scenes from KITTI. The experimental results show that the accuracy of the estimated poses is superior to other alternative methods. --- ### GaussianImage++: Boosted Image Representation and Compression with 2D Gaussian Splatting Authors: Tiantian Li, Xinjie Zhang, Xingtong Ge, Tongda Xu, Dailan He, Jun Zhang, Yan Wang Published: 2025-12-22 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.19108v2 Implicit neural representations (INRs) have achieved remarkable success in image representation and compression, but they require substantial training time and memory. Meanwhile, recent 2D Gaussian Splatting (GS) methods (\textit{e.g.}, GaussianImage) offer promising alternatives through efficient primitive-based rendering. However, these methods require excessive Gaussian primitives to maintain high visual fidelity. To exploit the potential of GS-based approaches, we present GaussianImage++, which utilizes limited Gaussian primitives to achieve impressive representation and compression performance. Firstly, we introduce a distortion-driven densification mechanism. It progressively allocates Gaussian primitives according to signal intensity. Secondly, we employ context-aware Gaussian filters for each primitive, which assist in the densification to optimize Gaussian primitives based on varying image content. Thirdly, we integrate attribute-separated learnable scalar quantizers and quantization-aware training, enabling efficient compression of primitive attributes. Experimental results demonstrate the effectiveness of our method. In particular, GaussianImage++ outperforms GaussianImage and INRs-based COIN in representation and compression performance while maintaining real-time decoding and low memory usage. --- ### FC-MIR: A Mobile Screen Awareness Framework for Intent-Aware Recommendation based on Frame-Compressed Multimodal Trajectory Reasoning Authors: Zhe Yang, Xiaoshuang Sheng, Zhengnan Zhang, Jidong Wu, Zexing Wang, Xin He, Shenghua Xu, Guanjing Xiong Published: 2025-12-22 Categories: cs.AI arXiv: https://arxiv.org/abs/2512.19107v1 Identifying user intent from mobile UI operation trajectories is critical for advancing UI understanding and enabling task automation agents. While Multimodal Large Language Models (MLLMs) excel at video understanding tasks, their real-time mobile deployment is constrained by heavy computational costs and inefficient redundant frame processing. To address these issues, we propose the FC-MIR framework: leveraging keyframe sampling and adaptive concatenation, it cuts visual redundancy to boost inference efficiency, while integrating state-of-the-art closed-source MLLMs or fine-tuned models (e.g., Qwen3-VL) for trajectory summarization and intent prediction. We further expand task scope to explore generating post-prediction operations and search suggestions, and introduce a fine-grained metric to evaluate the practical utility of summaries, predictions, and suggestions. For rigorous assessment, we construct a UI trajectory dataset covering scenarios from UI-Agents (Agent-I) and real user interactions (Person-I). Experimental results show our compression method retains performance at 50%-60% compression rates; both closed-source and fine-tuned MLLMs demonstrate strong intent summarization, supporting potential lightweight on-device deployment. However, MLLMs still struggle with useful and "surprising" suggestions, leaving room for improvement. Finally, we deploy the framework in a real-world setting, integrating UI perception and UI-Agent proxies to lay a foundation for future progress in this field. --- ### Explicit and Non-asymptotic Query Complexities of Rank-Based Zeroth-order Algorithm on Stochastic Smooth Functions Authors: Haishan Ye Published: 2025-12-22 Categories: math.OC, cs.LG arXiv: https://arxiv.org/abs/2512.19104v1 Zeroth-order (ZO) optimization with ordinal feedback has emerged as a fundamental problem in modern machine learning systems, particularly in human-in-the-loop settings such as reinforcement learning from human feedback, preference learning, and evolutionary strategies. While rank-based ZO algorithms enjoy strong empirical success and robustness properties, their theoretical understanding, especially under stochastic objectives and standard smoothness assumptions, remains limited. In this paper, we study rank-based zeroth-order optimization for stochastic functions where only ordinal feedback of the stochastic function is available. We propose a simple and computationally efficient rank-based ZO algorithm. Under standard assumptions including smoothness, strong convexity, and bounded second moments of stochastic gradients, we establish explicit non-asymptotic query complexity bounds for both convex and nonconvex objectives. Notably, our results match the best-known query complexities of value-based ZO algorithms, demonstrating that ordinal information alone is sufficient for optimal query efficiency in stochastic settings. Our analysis departs from existing drift-based and information-geometric techniques, offering new tools for the study of rank-based optimization under noise. These findings narrow the gap between theory and practice and provide a principled foundation for optimization driven by human preferences. --- ### Timely Parameter Updating in Over-the-Air Federated Learning Authors: Jiaqi Zhu, Zhongyuan Zhao, Xiao Li, Ruihao Du, Shi Jin, Howard H. Yang Published: 2025-12-22 Categories: cs.LG, cs.DC arXiv: https://arxiv.org/abs/2512.19103v1 Incorporating over-the-air computations (OAC) into the model training process of federated learning (FL) is an effective approach to alleviating the communication bottleneck in FL systems. Under OAC-FL, every client modulates its intermediate parameters, such as gradient, onto the same set of orthogonal waveforms and simultaneously transmits the radio signal to the edge server. By exploiting the superposition property of multiple-access channels, the edge server can obtain an automatically aggregated global gradient from the received signal. However, the limited number of orthogonal waveforms available in practical systems is fundamentally mismatched with the high dimensionality of modern deep learning models. To address this issue, we propose Freshness Freshness-mAgnItude awaRe top-k (FAIR-k), an algorithm that selects, in each communication round, the most impactful subset of gradients to be updated over the air. In essence, FAIR-k combines the complementary strengths of the Round-Robin and Top-k algorithms, striking a delicate balance between timeliness (freshness of parameter updates) and importance (gradient magnitude). Leveraging tools from Markov analysis, we characterize the distribution of parameter staleness under FAIR-k. Building on this, we establish the convergence rate of OAC-FL with FAIR-k, which discloses the joint effect of data heterogeneity, channel noise, and parameter staleness on the training efficiency. Notably, as opposed to conventional analyses that assume a universal Lipschitz constant across all the clients, our framework adopts a finer-grained model of the data heterogeneity. The analysis demonstrates that since FAIR-k promotes fresh (and fair) parameter updates, it not only accelerates convergence but also enhances communication efficiency by enabling an extended period of local training without significantly affecting overall training efficiency. --- ### Dual Model Deep Learning for Alzheimer Prognostication Authors: Alireza Moayedikia, Sara Fin, Uffe Kock Wiil Published: 2025-12-22 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.19099v1 Disease modifying therapies for Alzheimer's disease demand precise timing decisions, yet current predictive models require longitudinal observations and provide no uncertainty quantification, rendering them impractical at the critical first visit when treatment decisions must be made. We developed PROGRESS (PRognostic Generalization from REsting Static Signatures), a dual-model deep learning framework that transforms a single baseline cerebrospinal fluid biomarker assessment into actionable prognostic estimates without requiring prior clinical history. The framework addresses two complementary clinical questions: a probabilistic trajectory network predicts individualized cognitive decline with calibrated uncertainty bounds achieving near-nominal coverage, enabling honest prognostic communication; and a deep survival model estimates time to conversion from mild cognitive impairment to dementia. Using data from over 3,000 participants across 43 Alzheimer's Disease Research Centers in the National Alzheimer's Coordinating Center database, PROGRESS substantially outperforms Cox proportional hazards, Random Survival Forests, and gradient boosting methods for survival prediction. Risk stratification identifies patient groups with seven-fold differences in conversion rates, enabling clinically meaningful treatment prioritization. Leave-one-center-out validation demonstrates robust generalizability, with survival discrimination remaining strong across held-out sites despite heterogeneous measurement conditions spanning four decades of assay technologies. By combining superior survival prediction with trustworthy trajectory uncertainty quantification, PROGRESS bridges the gap between biomarker measurement and personalized clinical decision-making. --- ### DIVER-1: Scaling Intracranial EEG Foundation Models for Transferable Representations Authors: Danny Dongyeop Han, Yonghyeon Gwon, Ahhyun Lucy Lee, Taeyang Lee, Seong Jin Lee, Jubin Choi, Sebin Lee, Jihyun Bang, Seungju Lee, David Keetae Park, Shinjae Yoo, Chun Kee Chung et al. Published: 2025-12-22 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2512.19097v3 Intracranial EEG (iEEG) provides direct, millisecond-scale recordings of human neural activity, but reusable representation learning is difficult because electrode layouts, anatomical coverage, referencing schemes, and recording conditions vary across patients and centers. We introduce DIVER-1, a self-supervised iEEG foundation model for variable-input recordings that combines any-variate electrode-time attention, spatio-temporal resampling, input-conditioned positional embeddings, and multi-domain masked reconstruction without assuming a fixed electrode montage. We pretrain two variants, DIVER-1-0.1s and DIVER-1-1s, on 5,310 hours of ECoG and SEEG spanning 352k channel-hours, roughly 54x the BrainTreeBank-based pretraining volume. We evaluate DIVER-1 on two held-out benchmarks: Neuroprobe for naturalistic cognitive decoding and MAYO for seizure detection. On leakage-aware Neuroprobe, DIVER-1-0.1s outperforms prior evaluated iEEG foundation models despite using no BrainTreeBank recordings, the corpus underlying Neuroprobe, during pretraining; it also exceeds the linear spectrogram decoder in mean AUROC and remains competitive with stronger nonlinear baselines, a level prior evaluated iEEG foundation models did not reach. DIVER-1-1s also achieves the top AUROC on MAYO seizure detection. Finally, we conduct, to our knowledge, the first controlled compute-aware scaling study for self-supervised iEEG pretraining, sweeping data scale, subject count, training duration, and model size up to 1.8B parameters. Our results indicate a data-constrained regime: expanding unique recordings and training sufficiently long are more reliable scaling axes than increasing parameter count alone. Code is available at link. --- ### Conditioning Accept-Desirability models in the context of AGM-like belief change Authors: Kathelijne Coussement, Gert de Cooman, Keano De Vos Published: 2025-12-22 Categories: cs.AI, math.LO, math.PR arXiv: https://arxiv.org/abs/2512.19096v1 We discuss conditionalisation for Accept-Desirability models in an abstract decision-making framework, where uncertain rewards live in a general linear space, and events are special projection operators on that linear space. This abstract setting allows us to unify classical and quantum probabilities, and extend them to an imprecise probabilities context. We introduce a new conditioning rule for our Accept-Desirability models, based on the idea that observing an event introduces new indifferences between options. We associate a belief revision operator with our conditioning rule, and investigate which of the AGM axioms for belief revision still hold in our more general framework. We investigate two interesting special cases where all of these axioms are shown to still hold: classical propositional logic and full conditional probabilities. --- ### Mamba-Based Modality Disentanglement Network for Multi-Contrast MRI Reconstruction Authors: Weiyi Lyu, Xinming Fang, Jun Wang, Jun Shi, Guixu Zhang, Juncheng Li Published: 2025-12-22 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.19095v1 Magnetic resonance imaging (MRI) is a cornerstone of modern clinical diagnosis, offering unparalleled soft-tissue contrast without ionizing radiation. However, prolonged scan times remain a major barrier to patient throughput and comfort. Existing accelerated MRI techniques often struggle with two key challenges: (1) failure to effectively utilize inherent K-space prior information, leading to persistent aliasing artifacts from zero-filled inputs; and (2) contamination of target reconstruction quality by irrelevant information when employing multi-contrast fusion strategies. To overcome these challenges, we present MambaMDN, a dual-domain framework for multi-contrast MRI reconstruction. Our approach first employs fully-sampled reference K-space data to complete the undersampled target data, generating structurally aligned but modality-mixed inputs. Subsequently, we develop a Mamba-based modality disentanglement network to extract and remove reference-specific features from the mixed representation. Furthermore, we introduce an iterative refinement mechanism to progressively enhance reconstruction accuracy through repeated feature purification. Extensive experiments demonstrate that MambaMDN can significantly outperform existing multi-contrast reconstruction methods. --- ### Tool-Augmented Hybrid Ensemble Reasoning with Distillation for Bilingual Mathematical Problem Solving Authors: Peiqing Lu, Yuan Zhang, Haoyun Zhang, Jiasen Zheng, Kejian Tong, Wenjun Wu Published: 2025-12-22 Categories: cs.AI arXiv: https://arxiv.org/abs/2512.19093v1 Bilingual mathematical problem solving needs a clear link between language reasoning and symbolic calculation. Large language models often handle language well but are weak in accurate computation. This paper presents HERALD (Hybrid Ensemble Reasoning with Adaptive Learning and Distillation), a framework that joins reasoning and calculation using NuminaMath-7B-TIR, GPT-4o, and Mistral-7B. HERALD uses adaptive routing, tool-based reinforcement learning, and knowledge distillation to connect different reasoning paths. Confidence calibration keeps weighting stable, and dual-path checking keeps results correct. Reinforcement learning controls tool use to cut redundancy, and distillation lowers delay without hurting accuracy. The system shows that combining symbolic checking, adaptive ensembles, and bilingual fine-tuning helps achieve both fluent reasoning and precise calculation. HERALD offers a practical solution for multilingual mathematical reasoning with better accuracy, stability, and clarity. --- ### A Large Language Model Based Method for Complex Logical Reasoning over Knowledge Graphs Authors: Ziyan Zhang, Chao Wang, Zhuo Chen, Lei Chen, Chiyi Li, Kai Song Published: 2025-12-22 Categories: cs.CL arXiv: https://arxiv.org/abs/2512.19092v1 Reasoning over knowledge graphs (KGs) with first-order logic (FOL) queries is challenging due to the inherent incompleteness of real-world KGs and the compositional complexity of logical query structures. Most existing methods rely on embedding entities and relations into continuous geometric spaces and answer queries via differentiable set operations. While effective for simple query patterns, these approaches often struggle to generalize to complex queries involving multiple operators, deeper reasoning chains, or heterogeneous KG schemas. We propose ROG (Reasoning Over knowledge Graphs with large language models), an ensemble-style framework that combines query-aware KG neighborhood retrieval with large language model (LLM)-based chain-of-thought reasoning. ROG decomposes complex FOL queries into sequences of simpler sub-queries, retrieves compact, query-relevant subgraphs as contextual evidence, and performs step-by-step logical inference using an LLM, avoiding the need for task-specific embedding optimization. Experiments on standard KG reasoning benchmarks demonstrate that ROG consistently outperforms strong embedding-based baselines in terms of mean reciprocal rank (MRR), with particularly notable gains on high-complexity query types. These results suggest that integrating structured KG retrieval with LLM-driven logical reasoning offers a robust and effective alternative for complex KG reasoning tasks. --- ### Auditing Significance, Metric Choice, and Demographic Fairness in Medical AI Challenges Authors: Ariel Lubonja, Pedro R. A. S. Bassi, Wenxuan Li, Hualin Qiao, Randal Burns, Alan L. Yuille, Zongwei Zhou Published: 2025-12-22 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2512.19091v1 Open challenges have become the de facto standard for comparative ranking of medical AI methods. Despite their importance, medical AI leaderboards exhibit three persistent limitations: (1) score gaps are rarely tested for statistical significance, so rank stability is unknown; (2) single averaged metrics are applied to every organ, hiding clinically important boundary errors; (3) performance across intersecting demographics is seldom reported, masking fairness and equity gaps. We introduce RankInsight, an open-source toolkit that seeks to address these limitations. RankInsight (1) computes pair-wise significance maps that show the nnU-Net family outperforms Vision-Language and MONAI submissions with high statistical certainty; (2) recomputes leaderboards with organ-appropriate metrics, reversing the order of the top four models when Dice is replaced by NSD for tubular structures; and (3) audits intersectional fairness, revealing that more than half of the MONAI-based entries have the largest gender-race discrepancy on our proprietary Johns Hopkins Hospital dataset. The RankInsight toolkit is publicly released and can be directly applied to past, ongoing, and future challenges. It enables organizers and participants to publish rankings that are statistically sound, clinically meaningful, and demographically fair. --- ### Retrieving Objects from 3D Scenes with Box-Guided Open-Vocabulary Instance Segmentation Authors: Khanh Nguyen, Dasith de Silva Edirimuni, Ghulam Mubashar Hassan, Ajmal Mian Published: 2025-12-22 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.19088v1 Locating and retrieving objects from scene-level point clouds is a challenging problem with broad applications in robotics and augmented reality. This task is commonly formulated as open-vocabulary 3D instance segmentation. Although recent methods demonstrate strong performance, they depend heavily on SAM and CLIP to generate and classify 3D instance masks from images accompanying the point cloud, leading to substantial computational overhead and slow processing that limit their deployment in real-world settings. Open-YOLO 3D alleviates this issue by using a real-time 2D detector to classify class-agnostic masks produced directly from the point cloud by a pretrained 3D segmenter, eliminating the need for SAM and CLIP and significantly reducing inference time. However, Open-YOLO 3D often fails to generalize to object categories that appear infrequently in the 3D training data. In this paper, we propose a method that generates 3D instance masks for novel objects from RGB images guided by a 2D open-vocabulary detector. Our approach inherits the 2D detector's ability to recognize novel objects while maintaining efficient classification, enabling fast and accurate retrieval of rare instances from open-ended text queries. Our code will be made available at https://github.com/ndkhanh360/BoxOVIS. --- ### $γ(3,4)$ `Attention' in Cognitive Agents: Ontology-Free Knowledge Representations With Promise Theoretic Semantics Authors: Mark Burgess Published: 2025-12-22 Categories: cs.AI arXiv: https://arxiv.org/abs/2512.19084v1 The semantics and dynamics of `attention' are closely related to promise theoretic notions developed for autonomous agents and can thus easily be written down in promise framework. In this way one may establish a bridge between vectorized Machine Learning and Knowledge Graph representations without relying on language models implicitly. Our expectations for knowledge presume a degree of statistical stability, i.e. average invariance under repeated observation, or `trust' in the data. Both learning networks and knowledge graph representations can meaningfully coexist to preserve different aspects of data. While vectorized data are useful for probabilistic estimation, graphs preserve the intentionality of the source even under data fractionation. Using a Semantic Spacetime $γ(3,4)$ graph, one avoids complex ontologies in favour of classification of features by their roles in semantic processes. The latter favours an approach to reasoning under conditions of uncertainty. Appropriate attention to causal boundary conditions may lead to orders of magnitude compression of data required for such context determination, as required in the contexts of autonomous robotics, defence deployments, and ad hoc emergency services. --- ### Population-Evolve: a Parallel Sampling and Evolutionary Method for LLM Math Reasoning Authors: Yanzhi Zhang, Yitong Duan, Zhaoxi Zhang, Jiyan He, Shuxin Zheng Published: 2025-12-22 Categories: cs.AI arXiv: https://arxiv.org/abs/2512.19081v1 Test-time scaling has emerged as a promising direction for enhancing the reasoning capabilities of Large Language Models in last few years. In this work, we propose Population-Evolve, a training-free method inspired by Genetic Algorithms to optimize LLM reasoning. Our approach maintains a dynamic population of candidate solutions for each problem via parallel reasoning. By incorporating an evolve prompt, the LLM self-evolves its population in all iterations. Upon convergence, the final answer is derived via majority voting. Furthermore, we establish a unification framework that interprets existing test-time scaling strategies through the lens of genetic algorithms. Empirical results demonstrate that Population-Evolve achieves superior accuracy with low performance variance and computational efficiency. Our findings highlight the potential of evolutionary strategies to unlock the reasoning power of LLMs during inference. --- ### Watch Closely: Mitigating Object Hallucinations in Large Vision-Language Models with Disentangled Decoding Authors: Ruiqi Ma, Yu Yan, Chunhong Zhang, Minghao Yin, XinChao Liu, Zhihong Jin, Zheng Hu Published: 2025-12-22 Categories: cs.CV, cs.CL arXiv: https://arxiv.org/abs/2512.19070v1 Large Vision-Language Models (LVLMs) bridge the gap between visual and linguistic modalities, demonstrating strong potential across a variety of domains. However, despite significant progress, LVLMs still suffer from severe hallucination issues in object recognition tasks. These models often fail to accurately identify certain objects, leading to text generation that appears fluent but does not correspond to the visual content, which can have serious consequences in real-world applications. Recently, several methods have been proposed to alleviate LVLM hallucinations, but most focus solely on reducing hallucinations in the language modality. To mitigate hallucinations in both the language and visual modalities, we introduce Hallucination Disentangled Decoding (HDD) method that requires no training. HDD enhances the original image by segmenting it and selecting images that augment the original, while also utilizing a blank image to eliminate language prior hallucinations in both the original and segmented images. This design not only reduces the model's dependence on language priors but also enhances its visual performance. (Code: https://github.com/rickeyhhh/Hallucination-Disentangled-Decoding) --- ### Can abstract concepts from LLM improve SLM performance? Authors: Siddharth Tandon Published: 2025-12-22 Categories: cs.AI arXiv: https://arxiv.org/abs/2512.19069v1 Large language models (LLMs) excel at diverse tasks, but their deployment on resource-constrained devices remains challenging. Existing methods like quantization, pruning, and distillation can reduce memory footprint but often demand extensive experimentation and careful infrastructure design. Leveraging existing techniques for extracting high-level concepts (represented as steering vectors) from larger models, we investigate their transferability to smaller language models (SLM) during inference. We demonstrate through extensive experimentation that these concepts can be effectively transferred to smaller models, irrespective of their family (e.g., Phi, Llama, Qwen), leading to performance improvements across a wide range of tasks. Furthermore, we introduce inference-time scaling to enhance performance by dynamically adjusting the steering intensity which has resulted in a 7-15\% of accuracy improvement for Qwen3-0.6B. --- ### On Cost-Aware Designs for Sequential Hypothesis Testing Authors: George Vershinin, Asaf Cohen, Omer Gurewitz Published: 2025-12-22 Categories: cs.IT, cs.LG arXiv: https://arxiv.org/abs/2512.19067v3 We introduce Cost-Aware (CA) Sequential Hypothesis Testing (CASHT), in which an active decision-maker selects sensing actions with different, random costs to identify the true hypothesis under an average-error constraint $δ$, while minimizing the expected total cost (rather than the number of samples). For fixed costs, we prove that the optimal expected total cost scales as $Θ(\log(1/δ))$, and is achievable by Multihypothesis Sequential Probability Ratio Test-based procedures. We show that the CA design principle is to maximize the ratio of expected information gain to expected cost under the policy-induced action distribution. Guided by this principle, we adapt two classic policies to the CA setting and establish their asymptotic optimality. We then treat random costs under two revelation models: ex-post, where costs are disclosed only after a sample is obtained, and the cost-error tradeoff coincides with the fixed-cost case, and ex-ante, where costs accrue before acquisition, and the decision maker may cancel an action mid-operation. For the ex-ante model, we characterize when cancellation lowers the total cost and analyze several cost distributions in detail. Simulations confirm our findings that the CA variants consistently reduce total cost relative to their classic counterparts, and when action cancellation helps or hurts. --- ### Fraud Detection Through Large-Scale Graph Clustering with Heterogeneous Link Transformation Authors: Chi Liu Published: 2025-12-22 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2512.19061v1 Collaborative fraud, where multiple fraudulent accounts coordinate to exploit online payment systems, poses significant challenges due to the formation of complex network structures. Traditional detection methods that rely solely on high-confidence identity links suffer from limited coverage, while approaches using all available linkages often result in fragmented graphs with reduced clustering effectiveness. In this paper, we propose a novel graph-based fraud detection framework that addresses the challenge of large-scale heterogeneous graph clustering through a principled link transformation approach. Our method distinguishes between \emph{hard links} (high-confidence identity relationships such as phone numbers, credit cards, and national IDs) and \emph{soft links} (behavioral associations including device fingerprints, cookies, and IP addresses). We introduce a graph transformation technique that first identifies connected components via hard links, merges them into super-nodes, and then reconstructs a weighted soft-link graph amenable to efficient embedding and clustering. The transformed graph is processed using LINE (Large-scale Information Network Embedding) for representation learning, followed by HDBSCAN (Hierarchical Density-Based Spatial Clustering of Applications with Noise) for density-based cluster discovery. Experiments on a real-world payment platform dataset demonstrate that our approach achieves significant graph size reduction (from 25 million to 7.7 million nodes), doubles the detection coverage compared to hard-link-only baselines, and maintains high precision across identified fraud clusters. Our framework provides a scalable and practical solution for industrial-scale fraud detection systems. --- ### 6DAttack: Backdoor Attacks in the 6DoF Pose Estimation Authors: Jihui Guo, Zongmin Zhang, Zhen Sun, Yuhao Yang, Jinlin Wu, Fu Zhang, Xinlei He Published: 2025-12-22 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.19058v1 Deep learning advances have enabled accurate six-degree-of-freedom (6DoF) object pose estimation, widely used in robotics, AR/VR, and autonomous systems. However, backdoor attacks pose significant security risks. While most research focuses on 2D vision, 6DoF pose estimation remains largely unexplored. Unlike traditional backdoors that only change classes, 6DoF attacks must control continuous parameters like translation and rotation, rendering 2D methods inapplicable. We propose 6DAttack, a framework using 3D object triggers to induce controlled erroneous poses while maintaining normal behavior. Evaluations on PVNet, DenseFusion, and PoseDiffusion across LINEMOD, YCB-Video, and CO3D show high attack success rates (ASRs) without compromising clean performance. Backdoored models achieve up to 100% clean ADD accuracy and 100% ASR, with triggered samples reaching 97.70% ADD-P. Furthermore, a representative defense remains ineffective. Our findings reveal a serious, underexplored threat to 6DoF pose estimation. --- ### Efficient Personalization of Generative Models via Optimal Experimental Design Authors: Guy Schacht, Ziyad Sheebaelhamd, Riccardo De Santi, Mojmír Mutný, Andreas Krause Published: 2025-12-22 Categories: cs.LG, cs.IT arXiv: https://arxiv.org/abs/2512.19057v2 Preference learning from human feedback has the ability to align generative models with the needs of end-users. Human feedback is costly and time-consuming to obtain, which creates demand for data-efficient query selection methods. This work presents a novel approach that leverages optimal experimental design to ask humans the most informative preference queries, from which we can elucidate the latent reward function modeling user preferences efficiently. We formulate the problem of preference query selection as the one that maximizes the information about the underlying latent preference model. We show that this problem has a convex optimization formulation, and introduce a statistically and computationally efficient algorithm ED-PBRL that is supported by theoretical guarantees and can efficiently construct structured queries such as images or text. We empirically present the proposed framework by personalizing a text-to-image generative model to user-specific styles, showing that it requires less preference queries compared to random query selection. --- ### Decoupled Generative Modeling for Human-Object Interaction Synthesis Authors: Hwanhee Jung, Seunggwan Lee, Jeongyoon Yoon, SeungHyeon Kim, Giljoo Nam, Qixing Huang, Sangpil Kim Published: 2025-12-22 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.19049v2 Synthesizing realistic human-object interaction (HOI) is essential for 3D computer vision and robotics, underpinning animation and embodied control. Existing approaches often require manually specified intermediate waypoints and place all optimization objectives on a single network, which increases complexity, reduces flexibility, and leads to errors such as unsynchronized human and object motion or penetration. To address these issues, we propose Decoupled Generative Modeling for Human-Object Interaction Synthesis (DecHOI), which separates path planning and action synthesis. A trajectory generator first produces human and object trajectories without prescribed waypoints, and an action generator conditions on these paths to synthesize detailed motions. To further improve contact realism, we employ adversarial training with a discriminator that focuses on the dynamics of distal joints. The framework also models a moving counterpart and supports responsive, long-sequence planning in dynamic scenes, while preserving plan consistency. Across two benchmarks, FullBodyManipulation and 3D-FUTURE, DecHOI surpasses prior methods on most quantitative metrics and qualitative evaluations, and perceptual studies likewise prefer our results. --- ### WaTeRFlow: Watermark Temporal Robustness via Flow Consistency Authors: Utae Jeong, Sumin In, Hyunju Ryu, Jaewan Choi, Feng Yang, Jongheon Jeong, Seungryong Kim, Sangpil Kim Published: 2025-12-22 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.19048v1 Image watermarking supports authenticity and provenance, yet many schemes are still easy to bypass with various distortions and powerful generative edits. Deep learning-based watermarking has improved robustness to diffusion-based image editing, but a gap remains when a watermarked image is converted to video by image-to-video (I2V), in which per-frame watermark detection weakens. I2V has quickly advanced from short, jittery clips to multi-second, temporally coherent scenes, and it now serves not only content creation but also world-modeling and simulation workflows, making cross-modal watermark recovery crucial. We present WaTeRFlow, a framework tailored for robustness under I2V. It consists of (i) FUSE (Flow-guided Unified Synthesis Engine), which exposes the encoder-decoder to realistic distortions via instruction-driven edits and a fast video diffusion proxy during training, (ii) optical-flow warping with a Temporal Consistency Loss (TCL) that stabilizes per-frame predictions, and (iii) a semantic preservation loss that maintains the conditioning signal. Experiments across representative I2V models show accurate watermark recovery from frames, with higher first-frame and per-frame bit accuracy and resilience when various distortions are applied before or after video generation. --- ### Time-series Forecast for Indoor Zone Air Temperature with Long Horizons: A Case Study with Sensor-based Data from a Smart Building Authors: Liping Sun, Yucheng Guo, Siliang Lu, Zhenzhen Li Published: 2025-12-22 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.19038v2 With the press of global climate change, extreme weather and sudden weather changes are becoming increasingly common. To maintain a comfortable indoor environment and minimize the contribution of the building to climate change as much as possible, higher requirements are placed on the operation and control of HVAC systems, e.g., more energy-efficient and flexible to response to the rapid change of weather. This places demands on the rapid modeling and prediction of zone air temperatures of buildings. Compared to the traditional simulation-based approach such as EnergyPlus and DOE2, a hybrid approach combined physics and data-driven is more suitable. Recently, the availability of high-quality datasets and algorithmic breakthroughs have driven a considerable amount of work in this field. However, in the niche of short- and long-term predictions, there are still some gaps in existing research. This paper aims to develop a time series forecast model to predict the zone air temperature in a building located in America on a 2-week horizon. The findings could be further improved to support intelligent control and operation of HVAC systems (i.e. demand flexibility) and could also be used as hybrid building energy modeling. --- ### Elevating Intrusion Detection and Security Fortification in Intelligent Networks through Cutting-Edge Machine Learning Paradigms Authors: Md Minhazul Islam Munna, Md Mahbubur Rahman, Jaroslav Frnda, Muhammad Shahid Anwar, Alpamis Kutlimuratov Published: 2025-12-22 Categories: cs.CR, cs.LG arXiv: https://arxiv.org/abs/2512.19037v1 The proliferation of IoT devices and their reliance on Wi-Fi networks have introduced significant security vulnerabilities, particularly the KRACK and Kr00k attacks, which exploit weaknesses in WPA2 encryption to intercept and manipulate sensitive data. Traditional IDS using classifiers face challenges such as model overfitting, incomplete feature extraction, and high false positive rates, limiting their effectiveness in real-world deployments. To address these challenges, this study proposes a robust multiclass machine learning based intrusion detection framework. The methodology integrates advanced feature selection techniques to identify critical attributes, mitigating redundancy and enhancing detection accuracy. Two distinct ML architectures are implemented: a baseline classifier pipeline and a stacked ensemble model combining noise injection, Principal Component Analysis (PCA), and meta learning to improve generalization and reduce false positives. Evaluated on the AWID3 data set, the proposed ensemble architecture achieves superior performance, with an accuracy of 98%, precision of 98%, recall of 98%, and a false positive rate of just 2%, outperforming existing state-of-the-art methods. This work demonstrates the efficacy of combining preprocessing strategies with ensemble learning to fortify network security against sophisticated Wi-Fi attacks, offering a scalable and reliable solution for IoT environments. Future directions include real-time deployment and adversarial resilience testing to further enhance the model's adaptability. --- ### Distinguishing Visually Similar Actions: Prompt-Guided Semantic Prototype Modulation for Few-Shot Action Recognition Authors: Xiaoyang Li, Mingming Lu, Ruiqi Wang, Hao Li, Zewei Le Published: 2025-12-22 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.19036v1 Few-shot action recognition aims to enable models to quickly learn new action categories from limited labeled samples, addressing the challenge of data scarcity in real-world applications. Current research primarily addresses three core challenges: (1) temporal modeling, where models are prone to interference from irrelevant static background information and struggle to capture the essence of dynamic action features; (2) visual similarity, where categories with subtle visual differences are difficult to distinguish; and (3) the modality gap between visual-textual support prototypes and visual-only queries, which complicates alignment within a shared embedding space. To address these challenges, this paper proposes a CLIP-SPM framework, which includes three components: (1) the Hierarchical Synergistic Motion Refinement (HSMR) module, which aligns deep and shallow motion features to improve temporal modeling by reducing static background interference; (2) the Semantic Prototype Modulation (SPM) strategy, which generates query-relevant text prompts to bridge the modality gap and integrates them with visual features, enhancing the discriminability between similar actions; and (3) the Prototype-Anchor Dual Modulation (PADM) method, which refines support prototypes and aligns query features with a global semantic anchor, improving consistency across support and query samples. Comprehensive experiments across standard benchmarks, including Kinetics, SSv2-Full, SSv2-Small, UCF101, and HMDB51, demonstrate that our CLIP-SPM achieves competitive performance under 1-shot, 3-shot, and 5-shot settings. Extensive ablation studies and visual analyses further validate the effectiveness of each component and its contributions to addressing the core challenges. The source code and models are publicly available at GitHub. --- ### Automatic Neuronal Activity Segmentation in Fast Four Dimensional Spatio-Temporal Fluorescence Imaging using Bayesian Approach Authors: Ran Li, Pan Xiao, Kaushik Dutta, Youdong Guo Published: 2025-12-22 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.19032v1 Fluorescence Microcopy Calcium Imaging is a fundamental tool to in-vivo record and analyze large scale neuronal activities simultaneously at a single cell resolution. Automatic and precise detection of behaviorally relevant neuron activity from the recordings is critical to study the mapping of brain activity in organisms. However a perpetual bottleneck to this problem is the manual segmentation which is time and labor intensive and lacks generalizability. To this end, we present a Bayesian Deep Learning Framework to detect neuronal activities in 4D spatio-temporal data obtained by light sheet microscopy. Our approach accounts for the use of temporal information by calculating pixel wise correlation maps and combines it with spatial information given by the mean summary image. The Bayesian framework not only produces probability segmentation maps but also models the uncertainty pertaining to active neuron detection. To evaluate the accuracy of our framework we implemented the test of reproducibility to assert the generalization of the network to detect neuron activity. The network achieved a mean Dice Score of 0.81 relative to the synthetic Ground Truth obtained by Otsu's method and a mean Dice Score of 0.79 between the first and second run for test of reproducibility. Our method successfully deployed can be used for rapid detection of active neuronal activities for behavioural studies. --- ### A Surrogate-Augmented Symbolic CFD-Driven Training Framework for Accelerating Multi-objective Physical Model Development Authors: Yuan Fang, Fabian Waschkowski, Maximilian Reissmann, Richard D. Sandberg, Takuo Oda, Koichi Tanimoto Published: 2025-12-22 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.19031v1 Computational Fluid Dynamics (CFD)-driven training combines machine learning (ML) with CFD solvers to develop physically consistent closure models with improved predictive accuracy. In the original framework, each ML-generated candidate model is embedded in a CFD solver and evaluated against reference data, requiring hundreds to thousands of high-fidelity simulations and resulting in prohibitive computational cost for complex flows. To overcome this limitation, we propose an extended framework that integrates surrogate modeling into symbolic CFD-driven training in real time to reduce training cost. The surrogate model learns to approximate the errors of ML-generated models based on previous CFD evaluations and is continuously refined during training. Newly generated models are first assessed using the surrogate, and only those predicted to yield small errors or high uncertainty are subsequently evaluated with full CFD simulations. Discrete expressions generated by symbolic regression are mapped into a continuous space using averaged input-symbol values as inputs to a probabilistic surrogate model. To support multi-objective model training, particularly when fixed weighting of competing quantities is challenging, the surrogate is extended to a multi-output formulation by generalizing the kernel to a matrix form, providing one mean and variance prediction per training objective. Selection metrics based on these probabilistic outputs are used to identify an optimal training setup. The proposed surrogate-augmented CFD-driven training framework is demonstrated across a range of statistically one- and two-dimensional flows, including both single- and multi-expression model optimization. In all cases, the framework substantially reduces training cost while maintaining predictive accuracy comparable to that of the original CFD-driven approach. --- ### Recontextualization Mitigates Specification Gaming without Modifying the Specification Authors: Ariana Azarbal, Victor Gillioz, Vladimir Ivanov, Bryce Woodworth, Jacob Drori, Nevan Wichers, Aram Ebtekar, Alex Cloud, Alexander Matt Turner Published: 2025-12-22 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2512.19027v2 Developers often struggle to specify correct training labels and rewards. Perhaps they don't need to. We propose recontextualization, which reduces how often language models "game" training signals, performing misbehaviors those signals mistakenly reinforce. We show recontextualization prevents models from learning to 1) prioritize evaluation metrics over chat response quality; 2) special-case code to pass incorrect tests; 3) overwrite evaluation functions rather than write correct code; and 4) become sycophantic. Our method works by generating completions from prompts discouraging misbehavior and then recontextualizing them as though they were in response to prompts permitting misbehavior. Recontextualization trains language models to resist misbehavior even when instructions permit it. This mitigates the reinforcement of misbehavior from misspecified training signals, reducing specification gaming without improving the supervision signal. --- ### Finer-Personalization Rank: Fine-Grained Retrieval Examines Identity Preservation for Personalized Generation Authors: Connor Kilrain, David Carlyn, Julia Chae, Sara Beery, Wei-Lun Chao, Jianyang Gu Published: 2025-12-22 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2512.19026v1 The rise of personalized generative models raises a central question: how should we evaluate identity preservation? Given a reference image (e.g., one's pet), we expect the generated image to retain precise details attached to the subject's identity. However, current generative evaluation metrics emphasize the overall semantic similarity between the reference and the output, and overlook these fine-grained discriminative details. We introduce Finer-Personalization Rank, an evaluation protocol tailored to identity preservation. Instead of pairwise similarity, Finer-Personalization Rank adopts a ranking view: it treats each generated image as a query against an identity-labeled gallery consisting of visually similar real images. Retrieval metrics (e.g., mean average precision) measure performance, where higher scores indicate that identity-specific details (e.g., a distinctive head spot) are preserved. We assess identity at multiple granularities -- from fine-grained categories (e.g., bird species, car models) to individual instances (e.g., re-identification). Across CUB, Stanford Cars, and animal Re-ID benchmarks, Finer-Personalization Rank more faithfully reflects identity retention than semantic-only metrics and reveals substantial identity drift in several popular personalization methods. These results position the gallery-based protocol as a principled and practical evaluation for personalized generation. --- ### The Erasure Illusion: Stress-Testing the Generalization of LLM Forgetting Evaluation Authors: Hengrui Jia, Taoran Li, Jonas Guan, Varun Chandrasekaran Published: 2025-12-22 Categories: cs.CR, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2512.19025v2 Machine unlearning aims to remove specific data influences from trained models, a capability essential for adhering to copyright laws and ensuring AI safety. Current unlearning metrics typically measure success by monitoring the model's performance degradation on the specific unlearning dataset ($D_u$). We argue that for Large Language Models (LLMs), this evaluation paradigm is insufficient and potentially misleading. Many real-world uses of unlearning--motivated by copyright or safety--implicitly target not only verbatim content in $D_u$, but also behaviors influenced by the broader generalizations the model derived from it. We demonstrate that LLMs can pass standard unlearning evaluation and appear to have "forgotten" the target knowledge, while simultaneously retaining strong capabilities on content that is semantically adjacent to $D_u$. This phenomenon indicates that erasing exact sentences does not necessarily equate to removing the underlying knowledge. To address this gap, we propose Proximal Surrogate Generation (PSG), an automated stress-testing framework that generates a surrogate dataset, $\tilde{D}_u$. This surrogate set is constructed to be semantically derived from $D_u$ yet sufficiently distinct in embedding space. By comparing unlearning metric scores between $D_u$ and $\tilde{D}_u$, we can stress-test the reliability of the metric itself. Our extensive evaluation across three LLM families (Llama-3-8B, Qwen2.5-7B, and Zephyr-7B-$β$), three distinct datasets, and seven standard metrics reveals widespread inconsistencies. We find that current metrics frequently overestimate unlearning success, failing to detect retained knowledge exposed by our stress-test datasets. --- ### IndoorUAV: Benchmarking Vision-Language UAV Navigation in Continuous Indoor Environments Authors: Xu Liu, Yu Liu, Hanshuo Qiu, Yang Qirong, Zhouhui Lian Published: 2025-12-22 Categories: cs.RO, cs.AI arXiv: https://arxiv.org/abs/2512.19024v1 Vision-Language Navigation (VLN) enables agents to navigate in complex environments by following natural language instructions grounded in visual observations. Although most existing work has focused on ground-based robots or outdoor Unmanned Aerial Vehicles (UAVs), indoor UAV-based VLN remains underexplored, despite its relevance to real-world applications such as inspection, delivery, and search-and-rescue in confined spaces. To bridge this gap, we introduce \textbf{IndoorUAV}, a novel benchmark and method specifically tailored for VLN with indoor UAVs. We begin by curating over 1,000 diverse and structurally rich 3D indoor scenes from the Habitat simulator. Within these environments, we simulate realistic UAV flight dynamics to collect diverse 3D navigation trajectories manually, further enriched through data augmentation techniques. Furthermore, we design an automated annotation pipeline to generate natural language instructions of varying granularity for each trajectory. This process yields over 16,000 high-quality trajectories, comprising the \textbf{IndoorUAV-VLN} subset, which focuses on long-horizon VLN. To support short-horizon planning, we segment long trajectories into sub-trajectories by selecting semantically salient keyframes and regenerating concise instructions, forming the \textbf{IndoorUAV-VLA} subset. Finally, we introduce \textbf{IndoorUAV-Agent}, a novel navigation model designed for our benchmark, leveraging task decomposition and multimodal reasoning. We hope IndoorUAV serves as a valuable resource to advance research on vision-language embodied AI in the indoor aerial navigation domain. --- ### Steering Vision-Language Pre-trained Models for Incremental Face Presentation Attack Detection Authors: Haoze Li, Jie Zhang, Guoying Zhao, Stephen Lin, Shiguang Shan Published: 2025-12-22 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.19022v2 Face Presentation Attack Detection (PAD) demands incremental learning (IL) to combat evolving spoofing tactics and domains. Privacy regulations, however, forbid retaining past data, necessitating rehearsal-free IL (RF-IL). Vision-Language Pre-trained (VLP) models, with their prompt-tunable cross-modal representations, enable efficient adaptation to new spoofing styles and domains. Capitalizing on this strength, we propose \textbf{SVLP-IL}, a VLP-based RF-IL framework that balances stability and plasticity via \textit{Multi-Aspect Prompting} (MAP) and \textit{Selective Elastic Weight Consolidation} (SEWC). MAP isolates domain dependencies, enhances distribution-shift sensitivity, and mitigates forgetting by jointly exploiting universal and domain-specific cues. SEWC selectively preserves critical weights from previous tasks, retaining essential knowledge while allowing flexibility for new adaptations. Comprehensive experiments across multiple PAD benchmarks show that SVLP-IL significantly reduces catastrophic forgetting and enhances performance on unseen domains. SVLP-IL offers a privacy-compliant, practical solution for robust lifelong PAD deployment in RF-IL settings. --- ### VLNVerse: A Benchmark for Vision-Language Navigation with Versatile, Embodied, Realistic Simulation and Evaluation Authors: Sihao Lin, Zerui Li, Xunyi Zhao, Gengze Zhou, Liuyi Wang, Rong Wei, Rui Tang, Juncheng Li, Hanqing Wang, Jiangmiao Pang, Anton van den Hengel, Jiajun Liu et al. Published: 2025-12-22 Categories: cs.CV, cs.RO arXiv: https://arxiv.org/abs/2512.19021v1 Despite remarkable progress in Vision-Language Navigation (VLN), existing benchmarks remain confined to fixed, small-scale datasets with naive physical simulation. These shortcomings limit the insight that the benchmarks provide into sim-to-real generalization, and create a significant research gap. Furthermore, task fragmentation prevents unified/shared progress in the area, while limited data scales fail to meet the demands of modern LLM-based pretraining. To overcome these limitations, we introduce VLNVerse: a new large-scale, extensible benchmark designed for Versatile, Embodied, Realistic Simulation, and Evaluation. VLNVerse redefines VLN as a scalable, full-stack embodied AI problem. Its Versatile nature unifies previously fragmented tasks into a single framework and provides an extensible toolkit for researchers. Its Embodied design moves beyond intangible and teleporting "ghost" agents that support full-kinematics in a Realistic Simulation powered by a robust physics engine. We leverage the scale and diversity of VLNVerse to conduct a comprehensive Evaluation of existing methods, from classic models to MLLM-based agents. We also propose a novel unified multi-task model capable of addressing all tasks within the benchmark. VLNVerse aims to narrow the gap between simulated navigation and real-world generalization, providing the community with a vital tool to boost research towards scalable, general-purpose embodied locomotion agents. --- ### CETCAM: Camera-Controllable Video Generation via Consistent and Extensible Tokenization Authors: Zelin Zhao, Xinyu Gong, Bangya Liu, Ziyang Song, Jun Zhang, Suhui Wu, Yongxin Chen, Hao Zhang Published: 2025-12-22 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2512.19020v1 Achieving precise camera control in video generation remains challenging, as existing methods often rely on camera pose annotations that are difficult to scale to large and dynamic datasets and are frequently inconsistent with depth estimation, leading to train-test discrepancies. We introduce CETCAM, a camera-controllable video generation framework that eliminates the need for camera annotations through a consistent and extensible tokenization scheme. CETCAM leverages recent advances in geometry foundation models, such as VGGT, to estimate depth and camera parameters and converts them into unified, geometry-aware tokens. These tokens are seamlessly integrated into a pretrained video diffusion backbone via lightweight context blocks. Trained in two progressive stages, CETCAM first learns robust camera controllability from diverse raw video data and then refines fine-grained visual quality using curated high-fidelity datasets. Extensive experiments across multiple benchmarks demonstrate state-of-the-art geometric consistency, temporal stability, and visual realism. Moreover, CETCAM exhibits strong adaptability to additional control modalities, including inpainting and layout control, highlighting its flexibility beyond camera control. The project page is available at https://sjtuytc.github.io/CETCam_project_page.github.io/. --- ### Optimizer Dynamics at the Edge of Stability with Differential Privacy Authors: Ayana Hussain, Ricky Fang Published: 2025-12-22 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.19019v1 Deep learning models can reveal sensitive information about individual training examples, and while differential privacy (DP) provides guarantees restricting such leakage, it also alters optimization dynamics in poorly understood ways. We study the training dynamics of neural networks under DP by comparing Gradient Descent (GD), and Adam to their privacy-preserving variants. Prior work shows that these optimizers exhibit distinct stability dynamics: full-batch methods train at the Edge of Stability (EoS), while mini-batch and adaptive methods exhibit analogous edge-of-stability behavior. At these regimes, the training loss and the sharpness--the maximum eigenvalue of the training loss Hessian--exhibit certain characteristic behavior. In DP training, per-example gradient clipping and Gaussian noise modify the update rule, and it is unclear whether these stability patterns persist. We analyze how clipping and noise change sharpness and loss evolution and show that while DP generally reduces the sharpness and can prevent optimizers from fully reaching the classical stability thresholds, patterns from EoS and analogous adaptive methods stability regimes persist, with the largest learning rates and largest privacy budgets approaching, and sometimes exceeding, these thresholds. These findings highlight the unpredictability introduced by DP in neural network optimization. --- ### DramaBench: A Six-Dimensional Evaluation Framework for Drama Script Continuation Authors: Shijian Ma, Yunqi Huang, Yan Lin Published: 2025-12-22 Categories: cs.CL arXiv: https://arxiv.org/abs/2512.19012v2 Drama script continuation requires models to maintain character consistency, advance plot coherently, and preserve dramatic structurecapabilities that existing benchmarks fail to evaluate comprehensively. We present DramaBench, the first large-scale benchmark for evaluating drama script continuation across six independent dimensions: Format Standards, Narrative Efficiency, Character Consistency, Emotional Depth, Logic Consistency, and Conflict Handling. Our framework combines rulebased analysis with LLM-based labeling and statistical metrics, ensuring objective and reproducible evaluation. We conduct comprehensive evaluation of 8 state-of-the-art language models on 1,103 scripts (8,824 evaluations total), with rigorous statistical significance testing (252 pairwise comparisons, 65.9% significant) and human validation (188 scripts, substantial agreement on 3/5 dimensions). Our ablation studies confirm all six dimensions capture independent quality aspects (mean | r | = 0.020). DramaBench provides actionable, dimensionspecific feedback for model improvement and establishes a rigorous standard for creative writing evaluation. --- ### Do You Really Need a GPU to Guard Your LLM? CPU-Class Classifiers and Multi-Stage Pipelines for Safety Enforcement at Scale Authors: Vasudev Majhi, Dhruv Gupta, Advait Singh, Matthew Barker, Dhruv Kumar Published: 2025-12-22 Categories: cs.CR, cs.AI, cs.CL, cs.LG arXiv: https://arxiv.org/abs/2512.19011v3 Safety classifiers that screen LLM inputs for jailbreak attempts have become standard deployment components, yet almost all production systems rely on GPU-based models: fine-tuned transformers and LLM-as-a-judge pipelines. These approaches impose significant per-query latency and infrastructure cost. Very little research has asked whether CPU-based classifiers, such as support vector machines and gradient-boosted trees trained on TF-IDF features, can match their accuracy across the conditions that production deployments encounter. We evaluate five CPU classifier families, Mamba-130M as an SSM-based GPU classifier, and transformer-based GPU models (DeBERTa-v3 and Gemma-2B with LoRA) across nine jailbreak sources and three regimes: in-distribution (D1), out-of-distribution (D2), and adversarially obfuscated (D3). On D1, the best CPU classifier matches the best transformer GPU model at roughly one-fifth the deployment cost. On D2, CPU classifiers fail via confident miscalibration, producing high-confidence false negatives that bypass escalation entirely. On D3, CPU classifiers outperform transformer GPU models by more than 26 percentage points in F1. Based on these complementary failure modes, we design GuardChain, a three-stage safety pipeline (Regex -> CPU -> GPU) that routes each prompt to the cheapest stage capable of a confident decision. The CPU stage alone resolves 80\% of in-distribution prompts at near-peak accuracy, and the GPU stage recovers the out-of-distribution failures. For practitioners deploying LLM safety at scale, this work provides evidence that GPU-class infrastructure is unnecessary for the majority of traffic. --- ### The 6th International Verification of Neural Networks Competition (VNN-COMP 2025): Summary and Results Authors: Konstantin Kaulen, Tobias Ladner, Stanley Bak, Christopher Brix, Hai Duong, Thomas Flinkow, Taylor T. Johnson, Lukas Koller, Edoardo Manino, ThanhVu H Nguyen, Haoze Wu Published: 2025-12-22 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2512.19007v1 This report summarizes the 6th International Verification of Neural Networks Competition (VNN-COMP 2025), held as a part of the 8th International Symposium on AI Verification (SAIV), that was collocated with the 37th International Conference on Computer-Aided Verification (CAV). VNN-COMP is held annually to facilitate the fair and objective comparison of state-of-the-art neural network verification tools, encourage the standardization of tool interfaces, and bring together the neural network verification community. To this end, standardized formats for networks (ONNX) and specification (VNN-LIB) were defined, tools were evaluated on equal-cost hardware (using an automatic evaluation pipeline based on AWS instances), and tool parameters were chosen by the participants before the final test sets were made public. In the 2025 iteration, 8 teams participated on a diverse set of 16 regular and 9 extended benchmarks. This report summarizes the rules, benchmarks, participating tools, results, and lessons learned from this iteration of this competition. --- ### Context-Aware Initialization for Reducing Generative Path Length in Diffusion Language Models Authors: Tongyuan Miao, Gary Huang, Kai Jun Han, Annie Jiang Published: 2025-12-22 Categories: cs.CL, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2512.19004v1 Diffusion Large Language Models (DLLMs) enable fully parallel token decoding but often remain impractical at inference time due to the many denoising iterations required to refine an information-free, fully masked initialization into coherent text. Most existing acceleration methods focus on traversing this generative trajectory more efficiently via improved solvers or sampling strategies. We advance a complementary perspective: shorten the trajectory itself by starting closer to the target distribution through context-aware initialization. We propose a training-free interface that injects prompt-conditioned priors from a lightweight auxiliary model into the diffusion initialization, and instantiate it with two mechanisms: discrete token injection and representation-level embedding interpolation. Because injected priors can be imperfect and unmask-only decoding can over-commit early, we also introduce a simple confidence-based remasking mechanism as a form of prior skepticism. Preliminary evidence on GSM8K suggests that context-aware initialization can substantially reduce denoising iterations (about 35\% fewer function evaluations in our setting), while also exposing a key open challenge: naive warm-starting can degrade final accuracy relative to strong diffusion baselines. We use these findings to motivate a research agenda around calibration, revision mechanisms, and representation alignment for reliable warm-started diffusion decoding. --- ### ORPR: An OR-Guided Pretrain-then-Reinforce Learning Model for Inventory Management Authors: Lingjie Zhao, Xue Yu, Yongzhi Qi, Hao Hu, Jianshen Zhang, Yingzheng Ma, Shuyu Han, Wei Qi, Zuo-Jun Max Shen Published: 2025-12-22 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2512.19001v2 As the pursuit of synergy between Artificial Intelligence (AI) and Operations Research (OR) gains momentum in handling complex inventory systems, a critical challenge persists: how to effectively reconcile AI's adaptive perception with OR's structural rigor. To bridge this gap, we propose a novel OR-Guided "Pretrain-then-Reinforce" framework. To provide structured guidance, we propose a simulation-augmented OR model that generates high-quality reference decisions, implicitly capturing complex business constraints and managerial preferences. Leveraging these OR-derived decisions as foundational training labels, we design a domain-informed deep learning foundation model to establish foundational decision-making capabilities, followed by a reinforcement learning (RL) fine-tuning stage. Uniquely, we position RL as a deep alignment mechanism that enables the AI agent to internalize the optimality principles of OR, while simultaneously leveraging exploration for general policy refinement and allowing expert guidance for scenario-specific adaptation (e.g., promotional events). Validated through extensive numerical experiments and a field deployment at JD.com augmented by a Difference-in-Differences (DiD) analysis, our model significantly outperforms incumbent industrial practices, delivering real-world gains of a 5.27-day reduction in turnover and a 2.29% increase in in-stock rates, alongside a 29.95% decrease in holding costs. Contrary to the prevailing trend of brute-force model scaling, our study demonstrates that a lightweight, domain-informed model can deliver state-of-the-art performance and robust transferability when guided by structured OR logic. This approach offers a scalable and cost-effective paradigm for intelligent supply chain management, highlighting the value of deeply aligning AI with OR. --- ### Evaluating the Challenges of LLMs in Real-world Medical Follow-up: A Comparative Study and An Optimized Framework Authors: Jinyan Liu, Zikang Chen, Qinchuan Wang, Tan Xie, Heming Zheng, Xudong Lv Published: 2025-12-22 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2512.18999v1 When applied directly in an end-to-end manner to medical follow-up tasks, Large Language Models (LLMs) often suffer from uncontrolled dialog flow and inaccurate information extraction due to the complexity of follow-up forms. To address this limitation, we designed and compared two follow-up chatbot systems: an end-to-end LLM-based system (control group) and a modular pipeline with structured process control (experimental group). Experimental results show that while the end-to-end approach frequently fails on lengthy and complex forms, our modular method-built on task decomposition, semantic clustering, and flow management-substantially improves dialog stability and extraction accuracy. Moreover, it reduces the number of dialogue turns by 46.73% and lowers token consumption by 80% to 87.5%. These findings highlight the necessity of integrating external control mechanisms when deploying LLMs in high-stakes medical follow-up scenarios. --- ### Dual-Margin Embedding for Fine-Grained Long-Tailed Plant Taxonomy Authors: Cheng Yaw Low, Heejoon Koo, Jaewoo Park, Meeyoung Cha Published: 2025-12-22 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.18994v2 Taxonomic classification of ecological families, genera, and species underpins biodiversity monitoring and conservation. Existing computer vision methods typically address fine-grained recognition and long-tailed learning in isolation. However, additional challenges such as spatiotemporal domain shift, hierarchical taxonomic structure, and previously unseen taxa often co-occur in real-world deployment, leading to brittle performance under open-world conditions. We propose TaxoNet, an embedding learning framework with a theoretically grounded dual-margin objective that reshapes class decision boundaries under class imbalance to improve fine-grained discrimination while strengthening rare-class representation geometry. We evaluate TaxoNet in open-world settings that capture co-occurring recognition challenges. Leveraging diverse plant datasets, including Google Auto-Arborist (urban tree imagery), iNaturalist (Plantae observations across heterogeneous ecosystems), and NAFlora-Mini (herbarium collections), we demonstrate that TaxoNet consistently outperforms strong baselines, including multimodal foundation models. --- ### Training-Free Global Geometric Association for 4D LiDAR Panoptic Segmentation Authors: Gyeongrok Oh, Youngdong Jang, Jonghyun Choi, Suk-Ju Kang, Guang Lin, Sangpil Kim Published: 2025-12-22 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2512.18991v3 Dominant paradigms for 4D LiDAR panoptic segmentation are usually required to train deep neural networks with large superimposed point clouds or design dedicated modules for instance association. However, these approaches perform redundant point processing and consequently become computationally expensive, yet still overlook the rich geometric priors inherently provided by raw point clouds. To this end, we introduce \textsc{Geo-4D}, a simple yet effective training-free framework that unifies spatial and temporal reasoning, enabling holistic LiDAR perception over long time horizons. Specifically, we propose a global geometric association strategy that establishes consistent instance correspondences by estimating an optimal transformation between instance-level point sets. To mitigate instability caused by structural inconsistencies in point cloud observations, we propose a global geometry-aware soft matching mechanism that enforces spatially coherent point-wise correspondences grounded in the spatial distribution of instance point sets. Furthermore, our carefully designed pipeline, which considers three instance types-static, dynamic, and missing-offers computational efficiency and occlusion-aware matching. Our extensive experiments across both SemanticKITTI and nuScenes demonstrate that our method consistently outperforms state-of-the-art approaches, even without additional training or extra point cloud inputs. --- ### Affordance RAG: Hierarchical Multimodal Retrieval with Affordance-Aware Embodied Memory for Mobile Manipulation Authors: Ryosuke Korekata, Quanting Xie, Yonatan Bisk, Komei Sugiura Published: 2025-12-22 Categories: cs.RO, cs.CL, cs.CV arXiv: https://arxiv.org/abs/2512.18987v1 In this study, we address the problem of open-vocabulary mobile manipulation, where a robot is required to carry a wide range of objects to receptacles based on free-form natural language instructions. This task is challenging, as it involves understanding visual semantics and the affordance of manipulation actions. To tackle these challenges, we propose Affordance RAG, a zero-shot hierarchical multimodal retrieval framework that constructs Affordance-Aware Embodied Memory from pre-explored images. The model retrieves candidate targets based on regional and visual semantics and reranks them with affordance scores, allowing the robot to identify manipulation options that are likely to be executable in real-world environments. Our method outperformed existing approaches in retrieval performance for mobile manipulation instruction in large-scale indoor environments. Furthermore, in real-world experiments where the robot performed mobile manipulation in indoor environments based on free-form instructions, the proposed method achieved a task success rate of 85%, outperforming existing methods in both retrieval performance and overall task success. ---